Machine Learning using Image Labeler App, Classification Learner App

I have a image dataset which has defect data and good data (positive and negative dataset). I have used the Image labeler app for labeling defects and good bounding boxes for my entire dataset. I would like to apply HOG feature extraction on the defect and good labels from the gTruth obtained from the Image labelling session. I also would like to feed in these features extracted to the SVM using Classification learner app.

Answer 1

[Drew]

Here is an example of how to extract image features using a pretrained network, and then build a classifier with those features:

https://www.mathworks.com/help/deeplearning/ug/extract-image-features-using-pretrained-network.html

When extracting the features, you can substitute HOG instead of the pretrained network: https://www.mathworks.com/help/vision/ref/extracthogfeatures.html

When building the SVM classfier, you can substitute the Classification Learner App instead of using the MATLAB commandline.